
# This Python 3 environment comes with many helpful analytics libraries installed
# For example, here's several helpful packages to load

import numpy as np                   # for performing linear algebraic operations
import pandas as pd                  # for performing data processing, CSV file I/O (e.g. pd.read_csv)
from   pandas import DataFrame       # importing dataframe from pandas to work with dataframe
from   scipy.spatial import distance # to find the euclidean distances between arrays 
import warnings                      # to report the warning scenarios
from   sklearn.metrics import classification_report, confusion_matrix # to measure the quality of the model predictions
import matplotlib.pyplot as plt      # to work with plots
import seaborn as sns                # to work with statistical graphs
import plotly.express as px          # to work with matrix/ column oriented data
from sklearn.feature_selection import SelectKBest # Imported for feature selection technique
from sklearn.feature_selection import chi2        # Imported for feature selection technique
from sklearn.preprocessing import MinMaxScaler    # to convert the features to given range
from collections import Counter                   # to count the total number of values as dictionary
from imblearn.over_sampling import SMOTE,ADASYN   # used to distribute the data evenly over sampling technique
from sklearn.preprocessing import RobustScaler,StandardScaler,LabelEncoder,LabelBinarizer # imported for transforming the value
from sklearn.preprocessing import StandardScaler  # To perform standard scaling


# Reading and checking the shape of the data
data = pd.read_csv('fifa21.csv')
data.shape

(18541, 92)


data.head()


data.columns

Index(['Unnamed: 0', 'ID', 'Name', 'Age', 'Photo', 'Nationality', 'Flag',
       'Overall', 'Potential', 'Club', 'Club Logo', 'Value', 'Wage', 'Special',
       'Preferred Foot', 'Weak Foot', 'Skill Moves',
       'International Reputation', 'Work Rate', 'Body Type', 'Real Face',
       'Release Clause', 'Position', 'Jersey Number', 'Joined',
       'Contract Valid Until', 'Height', 'Weight', 'LS', 'ST', 'RS', 'LW',
       'LF', 'CF', 'RF', 'RW', 'LAM', 'CAM', 'RAM', 'LM', 'LCM', 'CM', 'RCM',
       'RM', 'LWB', 'LDM', 'CDM', 'RDM', 'RWB', 'LB', 'LCB', 'CB', 'RCB', 'RB',
       'GK', 'Likes', 'Dislikes', 'Following', 'Crossing', 'Finishing',
       'Heading Accuracy', 'Short Passing', 'Volleys', 'Dribbling', 'Curve',
       'FK Accuracy', 'Long Passing', 'Ball Control', 'Acceleration',
       'Sprint Speed', 'Agility', 'Reactions', 'Balance', 'Shot Power',
       'Jumping', 'Stamina', 'Strength', 'Long Shots', 'Aggression',
       'Interceptions', 'Positioning', 'Vision', 'Penalties', 'Composure',
       'Defensive Awareness', 'Standing Tackle', 'Sliding Tackle', 'GK Diving',
       'GK Handling', 'GK Kicking', 'GK Positioning', 'GK Reflexes'],
      dtype='object')


features = ['Name', 'Age', 'Nationality', 'Overall', 'Potential', 'Club', 'Value', 'Wage', 'Special',
       'Preferred Foot', 'Weak Foot', 'Skill Moves','International Reputation', 'Work Rate', 'Body Type', 
        'Position', 'Height', 'Weight', 'Likes', 'Dislikes', 'Following', 'Crossing', 'Finishing',
       'Heading Accuracy', 'Short Passing', 'Volleys', 'Dribbling', 'Curve','FK Accuracy', 'Long Passing', 
        'Ball Control', 'Acceleration','Sprint Speed', 'Agility', 'Reactions', 'Balance', 'Shot Power',
        'Jumping', 'Stamina', 'Strength', 'Long Shots', 'Aggression','Interceptions', 'Positioning', 
        'Vision', 'Penalties', 'Composure', 'Standing Tackle', 'Sliding Tackle', 
        'GK Diving','GK Handling', 'GK Kicking', 'GK Positioning', 'GK Reflexes']


# Incorporating those features in the dataframe
df = data[features]
df.shape

(18541, 54)


# Checking the column null values
df.isnull().sum()

Name                          0
Age                           0
Nationality                   0
Overall                       0
Potential                     0
Club                        371
Value                         0
Wage                          0
Special                       0
Preferred Foot                0
Weak Foot                     0
Skill Moves                   0
International Reputation      0
Work Rate                     0
Body Type                    51
Position                     25
Height                        0
Weight                        0
Likes                         0
Dislikes                      0
Following                     0
Crossing                    153
Finishing                   209
Heading Accuracy            133
Short Passing               220
Volleys                     163
Dribbling                   183
Curve                       120
FK Accuracy                  44
Long Passing                138
Ball Control                214
Acceleration                221
Sprint Speed                244
Agility                     226
Reactions                   210
Balance                     177
Shot Power                  116
Jumping                     204
Stamina                     257
Strength                    198
Long Shots                  154
Aggression                   95
Interceptions               150
Positioning                 166
Vision                      225
Penalties                    36
Composure                   652
Standing Tackle             130
Sliding Tackle              171
GK Diving                    30
GK Handling                  32
GK Kicking                   32
GK Positioning               34
GK Reflexes                  36
dtype: int64


# Histogram to see the distribution of the data
df.Volleys.hist()

<AxesSubplot:>


# Skew function to see how the data is skewed
print(df['Volleys'].skew())
df['Volleys'].describe()

-0.23510340268058694

count    18378.000000
mean        44.687997
std         17.746712
min          4.000000
25%         32.000000
50%         46.000000
75%         58.000000
max         90.000000
Name: Volleys, dtype: float64


# Skew function to see how the data is skewed
print(df['Balance'].skew())
df['Balance'].describe()

-0.6195177715778473

count    18364.000000
mean        64.528098
std         14.040552
min         17.000000
25%         57.000000
50%         67.000000
75%         74.000000
max         97.000000
Name: Balance, dtype: float64


# Inter Quartile Range check
Q1 = df.quantile(0.25)
Q3 = df.quantile(0.75)
IQR = Q3 - Q1
print(IQR)

Age                       8.0
Overall                   9.0
Potential                 8.0
Value               1850000.0
Wage                   9000.0
Special                 319.0
Likes                     6.0
Dislikes                  1.0
Following                33.0
Crossing                 25.0
Finishing                31.0
Heading Accuracy         20.0
Short Passing            14.0
Volleys                  26.0
Dribbling                17.0
Curve                    26.0
FK Accuracy              26.0
Long Passing             20.0
Ball Control             14.0
Acceleration             17.0
Sprint Speed             17.0
Agility                  18.0
Reactions                12.0
Balance                  17.0
Shot Power               20.0
Jumping                  15.0
Stamina                  17.0
Strength                 16.0
Long Shots               29.0
Aggression               25.0
Interceptions            38.5
Positioning              24.0
Vision                   19.0
Penalties                21.0
Composure                16.0
Standing Tackle          38.0
Sliding Tackle           39.0
GK Diving                 6.0
GK Handling               6.0
GK Kicking                6.0
GK Positioning            6.0
GK Reflexes               6.0
dtype: float64


df['Volleys'].replace({np.NaN:data['Volleys'].mean()},inplace=True)
df['Curve'].replace({np.NaN:data['Curve'].mean()},inplace=True)
df['Agility'].replace({np.NaN:data['Agility'].mean()},inplace=True)
df['Balance'].replace({np.NaN:data['Balance'].mean()},inplace=True)
df['Jumping'].replace({np.NaN:data['Jumping'].mean()},inplace=True)
df['Interceptions'].replace({np.NaN:data['Interceptions'].mean()},inplace=True)
df['Positioning'].replace({np.NaN:data['Positioning'].mean()},inplace=True)
df['Composure'].replace({np.NaN:data['Composure'].mean()},inplace=True)
df['Sliding Tackle'].replace({np.NaN:data['Sliding Tackle'].mean()},inplace=True)
df['Crossing'].replace({np.NaN:data['Crossing'].mean()},inplace=True)
df['Finishing'].replace({np.NaN:data['Finishing'].mean()},inplace=True)
df['Heading Accuracy'].replace({np.NaN:data['Heading Accuracy'].mean()},inplace=True)
df['Short Passing'].replace({np.NaN:data['Short Passing'].mean()},inplace=True)
df['Dribbling'].replace({np.NaN:data['Dribbling'].mean()},inplace=True)
df['FK Accuracy'].replace({np.NaN:data['FK Accuracy'].mean()},inplace=True)
df['Long Passing'].replace({np.NaN:data['Long Passing'].mean()},inplace=True)
df['Ball Control'].replace({np.NaN:data['Ball Control'].mean()},inplace=True)
df['Acceleration'].replace({np.NaN:data['Acceleration'].mean()},inplace=True)
df['Sprint Speed'].replace({np.NaN:data['Sprint Speed'].mean()},inplace=True)
df['Reactions'].replace({np.NaN:data['Reactions'].mean()},inplace=True)
df['Shot Power'].replace({np.NaN:data['Shot Power'].mean()},inplace=True)
df['Stamina'].replace({np.NaN:data['Stamina'].mean()},inplace=True)
df['Strength'].replace({np.NaN:data['Strength'].mean()},inplace=True)
df['Long Shots'].replace({np.NaN:data['Long Shots'].mean()},inplace=True)
df['Aggression'].replace({np.NaN:data['Aggression'].mean()},inplace=True)
df['Vision'].replace({np.NaN:data['Vision'].mean()},inplace=True)
df['Penalties'].replace({np.NaN:data['Penalties'].mean()},inplace=True)
df['Standing Tackle'].replace({np.NaN:data['Standing Tackle'].mean()},inplace=True)
df['GK Diving'].replace({np.NaN:data['GK Diving'].mean()},inplace=True)
df['GK Handling'].replace({np.NaN:data['GK Handling'].mean()},inplace=True)
df['GK Kicking'].replace({np.NaN:data['GK Kicking'].mean()},inplace=True)
df['GK Positioning'].replace({np.NaN:data['GK Positioning'].mean()},inplace=True)
df['GK Reflexes'].replace({np.NaN:data['GK Reflexes'].mean()},inplace=True)

C:\Users\91989\anaconda\lib\site-packages\pandas\core\generic.py:6619: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  return self._update_inplace(result)


nat_cnt=df.groupby('Nationality').apply(lambda x:x['Name'].count()).reset_index(name='Counts')
nat_cnt.sort_values(by='Counts',ascending=False,inplace=True)
top_20_nat_cnt=nat_cnt[:20]
fig=px.bar(top_20_nat_cnt,x='Nationality',y='Counts',color='Counts',title='Nationwise Representation in the FIFA Game')
fig.show()


cnt_best_avg=df.groupby('Nationality').apply(lambda x:np.average(x['Overall'])).reset_index(name='Overall Ratings')
cnt_best_cnt=df.groupby('Nationality').apply(lambda x:x['Overall'].count()).reset_index(name='Player Counts')
snt_best_avg_cnt=pd.merge(cnt_best_avg,cnt_best_cnt,how='inner',left_on='Nationality',right_on='Nationality')
sel_best_avg_cnt=snt_best_avg_cnt[snt_best_avg_cnt['Player Counts']>=200]
sel_best_avg_cnt.sort_values(by=['Overall Ratings','Player Counts'],ascending=[False,False])
px.scatter(sel_best_avg_cnt,x='Overall Ratings',y='Player Counts',color='Player Counts',size='Overall Ratings',hover_data=['Nationality'],title='Nationwise Player counts and Average Potential')


clb_cnt=df.groupby('Club').apply(lambda x:x['Name'].count()).reset_index(name='Counts')
clb_cnt.sort_values(by='Counts',ascending=False,inplace=True)
fig=px.scatter(clb_cnt,x='Club',y='Counts',color='Counts',title='Clubwise Player counts in FIFA 21')
fig.show()


cnt_best_avg=df.groupby('Club').apply(lambda x:np.average(x['Overall'])).reset_index(name='Overall Ratings')
cnt_best_cnt=df.groupby('Club').apply(lambda x:x['Overall'].count()).reset_index(name='Player Counts')
snt_best_avg_cnt=pd.merge(cnt_best_avg,cnt_best_cnt,how='inner',left_on='Club',right_on='Club')
sel_best_avg_cnt=snt_best_avg_cnt[snt_best_avg_cnt['Player Counts']>=25]
sel_best_avg_cnt.sort_values(by=['Overall Ratings','Player Counts'],ascending=[False,False])
px.scatter(sel_best_avg_cnt,x='Overall Ratings',y='Player Counts',color='Player Counts',size='Overall Ratings',hover_data=['Club'],title='Clubwise player counts and Average Potential')


props=df[['Name','Nationality','Club','Height','Weight']]
props['Ht in ft']=pd.to_numeric(props['Height'].str[0])
props['Ht in in']=pd.to_numeric(props['Height'].str.split("\'").str[1].str.strip('"'))
props['Ht in cm']=(props['Ht in ft']*12+props['Ht in in'])*2.54
props['Weight in lb']=pd.to_numeric(props['Weight'].str.strip('lbs'))
fig=px.scatter(props,x='Weight in lb',y='Ht in cm',color='Ht in cm',size='Weight in lb',hover_data=['Name','Nationality','Club'],title='Height vs Weight Variation of the players in FIFA 21')
fig.show()

<ipython-input-17-327b0aaca2bd>:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

<ipython-input-17-327b0aaca2bd>:3: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

<ipython-input-17-327b0aaca2bd>:4: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

<ipython-input-17-327b0aaca2bd>:5: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


pos_cnt=df.groupby('Position').apply(lambda x:x['Name'].count()).reset_index(name='Counts')
pos_cnt.sort_values(by='Counts',ascending=False,inplace=True)
top_20_pos_cnt=pos_cnt[:20]
fig=px.bar(top_20_pos_cnt,x='Position',y='Counts',color='Counts',title='Positionwise Player counts in FIFA 21')
fig.show()

# position abbreviation for reference

#  'Right Midfielder', 'Striker', 'Left Midfielder', 'Right Wing',
#        'Left Centre Midfielder', 'Right Safety', 'Running Back',
#        'Left Winger', 'Centre Midfielder', 'Cornerback',
#        'Centre Defensive Midfielder', 'Centre Attacking Midfielder',
#        'Linebacker', 'Right Attacking Midfielder',
#        'Right Centre Midfielder', 'Right Centre Back', 'Right Wing Back',
#        'Left Defensive Midfielder', 'Left Attacking Midfielder',
#        'Left Centre Back', 'Cente Forward', 'Long Snapper', 'Goalkeeper',
#        'Left Wing Back', 'Left Forward', 'Right Defensive Midfielder',
#        'Right Forward'


age_cnt=df.groupby('Age').apply(lambda x:x['Name'].count()).reset_index(name='Counts')
fig=px.bar(age_cnt,x='Age',y='Counts',color='Counts',title='Agewise Player distribution in FIFA 21')
fig.show()


top_play=df[['Name','Overall',"Age",'Club','Position']]
top_play.sort_values(by='Overall',ascending=False,inplace=True)
top_30_play=top_play[:100]
fig=px.scatter(top_30_play,x='Age',y='Overall',color='Age',size='Overall',hover_data=['Name','Club','Position'],title='Top Football Players in the FIFA 21 game')
fig.show()

C:\Users\91989\anaconda\lib\site-packages\pandas\util\_decorators.py:311: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


cond_1=df['Overall']!=df['Potential']
cond_2=df['Age']<25
df_fil=df[cond_1 & cond_2]
potential_play=df_fil[['Name','Age','Nationality','Club','Potential','Position','Overall','Value']]
potential_play.sort_values(by='Potential',ascending=False,inplace=True)
top_potential_play=potential_play[:50]
fig=px.scatter(potential_play,x='Age',y='Potential',size='Potential',color='Age',hover_data=['Name','Age','Nationality','Position','Overall','Value'],title='Age vs Maximum Potential Distribution of the young Players')
fig.show()


final_team=df_fil[['Name','Age','Potential','Position','Club']]
final_team.sort_values(by='Age',inplace=True)
Position_play=final_team.groupby('Position').apply(lambda x:np.max(x['Potential'])).reset_index(name='Potential')
player_pos=pd.merge(final_team,Position_play,how='inner',left_on=['Position','Potential'],right_on=['Position','Potential'])
Position_best=player_pos[['Name','Club','Age','Position','Potential']]
cm = sns.light_palette("black", as_cmap=True) 
Position_best.style.background_gradient(cmap=cm).set_precision(2)

<ipython-input-22-6d068811c047>:7: FutureWarning:

this method is deprecated in favour of `Styler.format(precision=..)`


india_team_all_players=df[df['Nationality']=='India']
india_team_2=india_team_all_players[['Name','Age','Overall','Position']]
india_team_2.sort_values(by='Position',inplace=True)
age_play=india_team_2.groupby('Position').apply(lambda x:np.max(x['Overall'])).reset_index(name='Overall')
player_pos=pd.merge(india_team_2,age_play,how='inner',left_on=['Position','Overall'],right_on=['Position','Overall'])
Position_best=player_pos[['Name','Age','Position','Overall']]
cm = sns.light_palette("cyan", as_cmap=True) 
Position_best.style.background_gradient(cmap=cm).set_precision(2)

<ipython-input-23-63178736c7f9>:8: FutureWarning:

this method is deprecated in favour of `Styler.format(precision=..)`


corr_ = df.corr()

f,ax = plt.subplots(figsize=(25, 10))
sns.heatmap(corr_,annot=True, linewidths=0.5, cmap="YlGnBu", fmt= '.1f',ax=ax)
plt.show()


# Dropping records not having value greater than zero to avoid null values
df_new = df[(df['Value'] > 0)]


# function for transforming each positions to particular main position into 4 major positions as forward, midfielder, defender and goalkeeper
def complex_function(vc):
  if vc in ['ST','RW','RS','LW','CF','LS','LF','RF']:
    return 'Forward'
  elif vc in ['RM','LM','LCM','CM','CAM','RAM','RCM','LDM','CDM','RDM','LAM']:
    return 'Midfielder'
  elif vc in ['RB','CB','LB','RCB','RWB','LCB','LWB']:
    return 'Defender'
  else:
    return 'Goalkeeper'


# Applying function to create new field based on certain transformation
df_new['Grouped_Position'] = df_new['Position'].apply(complex_function)

<ipython-input-17-49d2c1f136c3>:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


# Adding the new transformed position field as Y df and dropped position field and made that as x df
x = df_new.drop(['Position','Grouped_Position'], axis = 1)
y = df_new['Grouped_Position']

print("Shape of x :", x.shape)
print("Shape of y :", y.shape)

Shape of x : (18033, 53)
Shape of y : (18033,)


feature_name = list(x.columns)
# no of maximum features we need to select
num_feats=30


# Label Encoding to convert the features to model understandable format
from sklearn.preprocessing import LabelEncoder
le = LabelEncoder()
for i in range(0,x.shape[1]):
    if x.dtypes[i]=='object':
        x[x.columns[i]] = le.fit_transform(x[x.columns[i]])
        
print(x)

        Name  Age  Nationality  Overall  Potential  Club       Value  \
0       5283   18          162       71         87   598   4900000.0   
1       4574   32          162       86         86   588  35500000.0   
2       6299   17          161       68         87   116   1800000.0   
3      14368   23           21       81         85   755  23000000.0   
4       7594   28           34       82         82   280  22500000.0   
...      ...  ...          ...      ...        ...   ...         ...   
18535  14662   39           21       75         75   767    375000.0   
18536  15187   37           57       77         77   754    600000.0   
18537  15851   34           49       68         68   548    425000.0   
18539  16275   32           23       76         76    75   2200000.0   
18540  16303   35          149       67         67   389    110000.0   

           Wage  Special  Preferred Foot  ...     Vision  Penalties  \
0         500.0     1729               1  ...  63.000000       66.0   
1      150000.0     2144               1  ...  75.000000       85.0   
2        2000.0     1671               1  ...  65.000000       50.0   
3       50000.0     2071               0  ...  75.000000       73.0   
4      105000.0     2099               0  ...  83.000000       81.0   
...         ...      ...             ...  ...        ...        ...   
18535   10000.0     1730               0  ...  55.119404       83.0   
18536   15000.0     1086               1  ...  31.000000       23.0   
18537    6000.0     1744               1  ...  68.000000       64.0   
18539   20000.0     1920               1  ...  76.000000       67.0   
18540    6000.0     1413               0  ...  40.000000       51.0   

       Composure  Standing Tackle  Sliding Tackle  GK Diving  GK Handling  \
0      61.000000             11.0            18.0        9.0         12.0   
1      80.000000             48.0            39.0       12.0          5.0   
2      59.000000             23.0            24.0       10.0         13.0   
3      79.000000             54.0            38.0        8.0          7.0   
4      87.000000             41.0            44.0       15.0         15.0   
...          ...              ...             ...        ...          ...   
18535  59.603052             23.0            24.0       11.0          9.0   
18536  20.000000             11.0            11.0       76.0         76.0   
18537  70.000000             27.0            33.0       14.0          9.0   
18539  59.603052             62.0            70.0        9.0          9.0   
18540  59.603052             65.0            69.0       13.0         15.0   

       GK Kicking  GK Positioning  GK Reflexes  
0             7.0             8.0          7.0  
1            13.0            13.0         10.0  
2            14.0            12.0          7.0  
3            13.0             8.0         14.0  
4            15.0             5.0         14.0  
...           ...             ...          ...  
18535        14.0            10.0         11.0  
18536        77.0            77.0         77.0  
18537        14.0            11.0         12.0  
18539         7.0             9.0         15.0  
18540         9.0             5.0          5.0  

[18033 rows x 53 columns]


# converted the features to given range and selected the top 30 features
x_norm = MinMaxScaler().fit_transform(x)
chi_selector = SelectKBest(chi2, k=num_feats)
chi_selector.fit(x_norm, y)
chi_support = chi_selector.get_support()
chi_feature = x.loc[:,chi_support].columns.tolist()
print(str(len(chi_feature)), 'selected features')

30 selected features


# Displaying the top 30 features
chi_feature

['Special',
 'Skill Moves',
 'Crossing',
 'Finishing',
 'Heading Accuracy',
 'Short Passing',
 'Volleys',
 'Dribbling',
 'Curve',
 'FK Accuracy',
 'Long Passing',
 'Ball Control',
 'Acceleration',
 'Sprint Speed',
 'Agility',
 'Balance',
 'Stamina',
 'Long Shots',
 'Aggression',
 'Interceptions',
 'Positioning',
 'Vision',
 'Penalties',
 'Standing Tackle',
 'Sliding Tackle',
 'GK Diving',
 'GK Handling',
 'GK Kicking',
 'GK Positioning',
 'GK Reflexes']


y = df_new['Grouped_Position']
x = df_new.drop(['Age', 'Nationality', 'Overall', 'Potential', 'Club', 'Value', 'Wage', 'Position', 'Grouped_Position', 'Weak Foot', 'International Reputation', 'Work Rate', 'Body Type', 'Weight', 'Likes', 'Dislikes', 'Following', 'Reactions', 'Balance', 'Shot Power', 'Jumping', 'Strength', 'Vision', 'Composure'], axis = 1)

print("Shape of x :", x.shape)
print("Shape of y :", y.shape)

Shape of x : (18033, 31)
Shape of y : (18033,)


# Label Encoding for final dataframe
from sklearn.preprocessing import LabelEncoder
le = LabelEncoder()
for i in range(0,x.shape[1]):
    if x.dtypes[i]=='object':
        x[x.columns[i]] = le.fit_transform(x[x.columns[i]])
        
print(x)

        Name  Special  Preferred Foot  Skill Moves  Height   Crossing  \
0       5283     1729               1            3      10  67.000000   
1       4574     2144               1            3      12  68.000000   
2       6299     1671               1            6      12  54.000000   
3      14368     2071               0            6      10  76.000000   
4       7594     2099               0            6       2  90.000000   
...      ...      ...             ...          ...     ...        ...   
18535  14662     1730               0            8      12  51.262508   
18536  15187     1086               1            0      12  10.000000   
18537  15851     1744               1            3       6  69.000000   
18539  16275     1920               1            3       2  61.000000   
18540  16303     1413               0            1      12  44.000000   

       Finishing  Heading Accuracy  Short Passing    Volleys  ...  \
0       60.00000              50.0       67.00000  56.000000  ...   
1       87.00000              89.0       75.00000  88.000000  ...   
2       59.00000              42.0       73.00000  59.000000  ...   
3       72.00000              45.0       74.00000  72.000000  ...   
4       84.00000              62.0       83.00000  90.000000  ...   
...          ...               ...            ...        ...  ...   
18535   47.51451              64.0       60.13258  44.687997  ...   
18536   10.00000              10.0       24.00000  10.000000  ...   
18537   61.00000              36.0       65.00000  70.000000  ...   
18539   65.00000              70.0       83.00000  73.000000  ...   
18540   21.00000              75.0       55.00000  23.000000  ...   

       Interceptions  Positioning  Penalties  Standing Tackle  Sliding Tackle  \
0               31.0    65.000000       66.0             11.0            18.0   
1               54.0    92.000000       85.0             48.0            39.0   
2               31.0    56.000000       50.0             23.0            24.0   
3               55.0    81.000000       73.0             54.0            38.0   
4               55.0    80.000000       81.0             41.0            44.0   
...              ...          ...        ...              ...             ...   
18535           21.0    51.943673       83.0             23.0            24.0   
18536           27.0    16.000000       23.0             11.0            11.0   
18537           39.0    66.000000       64.0             27.0            33.0   
18539           82.0    65.000000       67.0             62.0            70.0   
18540           65.0    39.000000       51.0             65.0            69.0   

       GK Diving  GK Handling  GK Kicking  GK Positioning  GK Reflexes  
0            9.0         12.0         7.0             8.0          7.0  
1           12.0          5.0        13.0            13.0         10.0  
2           10.0         13.0        14.0            12.0          7.0  
3            8.0          7.0        13.0             8.0         14.0  
4           15.0         15.0        15.0             5.0         14.0  
...          ...          ...         ...             ...          ...  
18535       11.0          9.0        14.0            10.0         11.0  
18536       76.0         76.0        77.0            77.0         77.0  
18537       14.0          9.0        14.0            11.0         12.0  
18539        9.0          9.0         7.0             9.0         15.0  
18540       13.0         15.0         9.0             5.0          5.0  

[18033 rows x 31 columns]


# using the train test split validation approach
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(x, y, test_size = 0.3, random_state = 0)

print("Shape of X_train :", X_train.shape)
print("Shape of X_test :", X_test.shape)
print("Shape of y_train :", y_train.shape)
print("Shape of y_test :", y_test.shape)

Shape of X_train : (12623, 31)
Shape of X_test : (5410, 31)
Shape of y_train : (12623,)
Shape of y_test : (5410,)


# the counts of the target variable to check the data imbalance
print('Classes and number of values in trainset',Counter(y_train))

Classes and number of values in trainset Counter({'Midfielder': 4743, 'Defender': 4022, 'Forward': 2663, 'Goalkeeper': 1195})


# Adaptive Synthetic Sampling Approach used for saampling the data
oversample = ADASYN(sampling_strategy='minority')
X_train,y_train = oversample.fit_resample(X_train,y_train)
print('Classes and number of values in trainset after ADSYN:',Counter(y_train))

Classes and number of values in trainset after ADSYN: Counter({'Midfielder': 4743, 'Goalkeeper': 4739, 'Defender': 4022, 'Forward': 2663})


# Standard Scaling technique applied to the train and test dataframe
sc = StandardScaler()
X_train = sc.fit_transform(X_train)
X_test = sc.transform(X_test)


from sklearn.svm import SVC
from sklearn.model_selection import RandomizedSearchCV
from sklearn.model_selection import KFold
from sklearn.model_selection import cross_val_score
svc = SVC( kernel= 'rbf', gamma= 0.1, C= 9)
svc.fit(X_train, y_train)
y_pred = svc.predict(X_test)
# print('Training Accuracy: %1.3f.' % svc.score(X_train, y_train))
#10-fold cross validation score
cv = cross_val_score(estimator = svc, X = X_train, y = y_train, cv =3)
SVC_cv = cv.mean()
print("10 fold cross validation :", SVC_cv)
from sklearn.metrics import accuracy_score
SVC_ac = accuracy_score(y_pred,y_test)
print("accuracy :", SVC_ac)
from sklearn.metrics import precision_score
SVC_p = precision_score(y_pred,y_test,average='weighted')
print("precision :", SVC_p)
from sklearn.metrics import recall_score
SVC_r = recall_score(y_pred,y_test,average='weighted')
print("recall :", SVC_r)
from sklearn.metrics import f1_score
SVC_f1 = f1_score(y_test, y_pred,average='weighted')
print("F1 Score :", SVC_f1)

10 fold cross validation : 0.8909507020473805
accuracy : 0.8604436229205176
precision : 0.8610622371180854
recall : 0.8604436229205176
F1 Score : 0.8602794204310344


svc_params = {'C': range(1, 10, 1), 'gamma': np.arange(0.1, 1, 0.1), 'kernel': ['rbf', 'linear']}


random_search = RandomizedSearchCV(estimator = svc, param_distributions = svc_params, n_iter = 20, cv = 5, verbose=2, random_state=42, n_jobs = -1)
random_search.fit(X_train, y_train)

Fitting 5 folds for each of 20 candidates, totalling 100 fits

RandomizedSearchCV(cv=5, estimator=SVC(C=9, gamma=0.1), n_iter=20, n_jobs=-1,
                   param_distributions={'C': range(1, 10),
                                        'gamma': array([0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9]),
                                        'kernel': ['rbf', 'linear']},
                   random_state=42, verbose=2)


print('\n All results:')
print(random_search.cv_results_)
print('\n Best estimator:')
print(random_search.best_estimator_)
print('\n Best hyperparameters:')
print(random_search.best_params_)
results = pd.DataFrame(random_search.cv_results_)

 All results:
{'mean_fit_time': array([68.2390811 , 15.48659596, 19.58205657, 13.26078887, 27.30172582,
       10.9647934 , 15.49956112, 11.71959081, 49.56279974, 11.19839506,
       10.26906362, 60.39004378,  8.69066701, 61.01321259, 38.1889441 ,
       61.68121591, 18.96536489, 69.70167737, 46.55615177, 67.94903998]), 'std_fit_time': array([2.16893897, 2.08876823, 0.81166038, 0.75011596, 0.98680485,
       0.25964075, 0.50468244, 0.11052728, 1.95869859, 0.68172382,
       0.31046006, 4.39795038, 0.11590589, 3.70234192, 1.048439  ,
       4.44608852, 0.56081856, 2.10189924, 2.0910459 , 1.01146143]), 'mean_score_time': array([15.12459397,  1.31876559,  1.39826078,  1.30595975,  9.20751724,
        1.23598275,  1.332445  ,  1.23233213, 13.77409987,  5.77425814,
        1.36029897, 15.14016943,  1.34795499, 15.18642173, 11.90248518,
       15.37271132,  1.35440249, 15.76266313, 13.31543345, 14.29841886]), 'std_score_time': array([0.07841032, 0.1367208 , 0.0847233 , 0.14839956, 0.58882183,
       0.04765614, 0.0905552 , 0.10572927, 0.4286374 , 0.6298421 ,
       0.10532052, 0.89086126, 0.03448069, 0.37851863, 0.14915417,
       0.52726209, 0.0773279 , 0.71106932, 1.44082216, 0.74783377]), 'param_kernel': masked_array(data=['rbf', 'linear', 'linear', 'linear', 'rbf', 'linear',
                   'linear', 'linear', 'rbf', 'rbf', 'linear', 'rbf',
                   'linear', 'rbf', 'rbf', 'rbf', 'linear', 'rbf', 'rbf',
                   'rbf'],
             mask=[False, False, False, False, False, False, False, False,
                   False, False, False, False, False, False, False, False,
                   False, False, False, False],
       fill_value='?',
            dtype=object), 'param_gamma': masked_array(data=[0.8, 0.1, 0.30000000000000004, 0.1,
                   0.30000000000000004, 0.6, 0.6, 0.8, 0.6, 0.1, 0.1,
                   0.7000000000000001, 0.8, 0.7000000000000001, 0.4,
                   0.7000000000000001, 0.9, 0.8, 0.5, 0.9],
             mask=[False, False, False, False, False, False, False, False,
                   False, False, False, False, False, False, False, False,
                   False, False, False, False],
       fill_value='?',
            dtype=object), 'param_C': masked_array(data=[9, 7, 8, 4, 6, 2, 6, 3, 6, 9, 2, 5, 1, 4, 2, 2, 8, 6,
                   6, 1],
             mask=[False, False, False, False, False, False, False, False,
                   False, False, False, False, False, False, False, False,
                   False, False, False, False],
       fill_value='?',
            dtype=object), 'params': [{'kernel': 'rbf', 'gamma': 0.8, 'C': 9}, {'kernel': 'linear', 'gamma': 0.1, 'C': 7}, {'kernel': 'linear', 'gamma': 0.30000000000000004, 'C': 8}, {'kernel': 'linear', 'gamma': 0.1, 'C': 4}, {'kernel': 'rbf', 'gamma': 0.30000000000000004, 'C': 6}, {'kernel': 'linear', 'gamma': 0.6, 'C': 2}, {'kernel': 'linear', 'gamma': 0.6, 'C': 6}, {'kernel': 'linear', 'gamma': 0.8, 'C': 3}, {'kernel': 'rbf', 'gamma': 0.6, 'C': 6}, {'kernel': 'rbf', 'gamma': 0.1, 'C': 9}, {'kernel': 'linear', 'gamma': 0.1, 'C': 2}, {'kernel': 'rbf', 'gamma': 0.7000000000000001, 'C': 5}, {'kernel': 'linear', 'gamma': 0.8, 'C': 1}, {'kernel': 'rbf', 'gamma': 0.7000000000000001, 'C': 4}, {'kernel': 'rbf', 'gamma': 0.4, 'C': 2}, {'kernel': 'rbf', 'gamma': 0.7000000000000001, 'C': 2}, {'kernel': 'linear', 'gamma': 0.9, 'C': 8}, {'kernel': 'rbf', 'gamma': 0.8, 'C': 6}, {'kernel': 'rbf', 'gamma': 0.5, 'C': 6}, {'kernel': 'rbf', 'gamma': 0.9, 'C': 1}], 'split0_test_score': array([0.76530612, 0.90445269, 0.90445269, 0.90414348, 0.88033395,
       0.9047619 , 0.90414348, 0.9047619 , 0.82653061, 0.89888683,
       0.9047619 , 0.79746444, 0.9047619 , 0.79746444, 0.8729128 ,
       0.79931973, 0.90445269, 0.76530612, 0.84662956, 0.70717378]), 'split1_test_score': array([0.80952381, 0.89424861, 0.89393939, 0.89332096, 0.88435374,
       0.89332096, 0.89424861, 0.89332096, 0.85312307, 0.89703154,
       0.89332096, 0.82869511, 0.89393939, 0.82869511, 0.88651824,
       0.82807669, 0.89393939, 0.80952381, 0.87167594, 0.77489177]), 'split2_test_score': array([0.81596041, 0.89266935, 0.89266935, 0.89266935, 0.88400866,
       0.89266935, 0.89266935, 0.89266935, 0.85988246, 0.89359728,
       0.89266935, 0.83884937, 0.89266935, 0.83884937, 0.88277142,
       0.84008661, 0.89266935, 0.81596041, 0.87349211, 0.76616146]), 'split3_test_score': array([0.83606557, 0.89514383, 0.89452521, 0.89514383, 0.88277142,
       0.89514383, 0.89514383, 0.89483452, 0.86792453, 0.89174142,
       0.89514383, 0.85307764, 0.89483452, 0.85307764, 0.88864831,
       0.85276833, 0.89452521, 0.83606557, 0.87967832, 0.79616455]), 'split4_test_score': array([0.8004949 , 0.90504176, 0.90504176, 0.90504176, 0.88617383,
       0.90473245, 0.90504176, 0.90504176, 0.8580266 , 0.89607176,
       0.90473245, 0.83235385, 0.90504176, 0.83235385, 0.888339  ,
       0.83266316, 0.90504176, 0.8004949 , 0.87349211, 0.75409836]), 'mean_test_score': array([0.80547016, 0.89831125, 0.89812568, 0.89806387, 0.88352832,
       0.8981257 , 0.8982494 , 0.8981257 , 0.85309745, 0.89546576,
       0.8981257 , 0.83008808, 0.89824938, 0.83008808, 0.88383796,
       0.8305829 , 0.89812568, 0.80547016, 0.86899361, 0.75969799]), 'std_test_score': array([0.02323988, 0.00531763, 0.00544285, 0.00539954, 0.00193327,
       0.00546694, 0.00524717, 0.00557782, 0.01411363, 0.00252525,
       0.00546694, 0.01831205, 0.00547583, 0.01831205, 0.00584943,
       0.01772291, 0.00544285, 0.02323988, 0.01150636, 0.02963979]), 'rank_test_score': array([18,  1,  7,  9, 12,  4,  2,  4, 14, 10,  4, 16,  3, 16, 11, 15,  7,
       18, 13, 20])}

 Best estimator:
SVC(C=7, gamma=0.1, kernel='linear')

 Best hyperparameters:
{'kernel': 'linear', 'gamma': 0.1, 'C': 7}


from sklearn.svm import SVC
from sklearn.model_selection import RandomizedSearchCV
from sklearn.model_selection import KFold
from sklearn.model_selection import cross_val_score
tuned_svc = SVC( kernel= 'linear', gamma= 0.1, C= 7)
tuned_svc.fit(X_train, y_train)
y_pred = tuned_svc.predict(X_test)
#10-fold cross validation score
cv = cross_val_score(estimator = svc, X = X_train, y = y_train, cv =3)
tuned_SVC_cv = cv.mean()
print("10 fold cross validation :", SVC_cv)
from sklearn.metrics import accuracy_score
tuned_SVC_ac = accuracy_score(y_pred,y_test)
print("accuracy :", SVC_ac)
from sklearn.metrics import precision_score
tuned_SVC_p = precision_score(y_pred,y_test,average='weighted')
print("precision :", SVC_p)
from sklearn.metrics import recall_score
tuned_SVC_r = recall_score(y_pred,y_test,average='weighted')
print("recall :", SVC_r)
from sklearn.metrics import f1_score
tuned_SVC_f1 = f1_score(y_test, y_pred,average='weighted')
print("F1 Score :", SVC_f1)

10 fold cross validation : 0.8909507020473805
accuracy : 0.8604436229205176
precision : 0.8610622371180854
recall : 0.8604436229205176
F1 Score : 0.8602794204310344


from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import KFold
from sklearn.model_selection import cross_val_score
lr = LogisticRegression(penalty = 'l2', C = 0.01 )
lr.fit(X_train, y_train)
y_pred = lr.predict(X_test)
#10-fold cross validation score
cv = cross_val_score(estimator = lr, X = X_train, y = y_train, cv =3)
log_cv = cv.mean()
print("10 fold cross validation :", log_cv)
from sklearn.metrics import accuracy_score
log_ac = accuracy_score(y_pred,y_test)
print("accuracy :", log_ac)
from sklearn.metrics import precision_score
log_p = precision_score(y_pred,y_test,average='weighted')
print("precision :", log_p)
from sklearn.metrics import recall_score
log_r = recall_score(y_pred,y_test,average='weighted')
print("recall :", log_r)
from sklearn.metrics import f1_score
log_f1 = f1_score(y_test, y_pred,average='weighted')
print("F1 Score :", log_f1)

10 fold cross validation : 0.8915073916001731
accuracy : 0.865988909426987
precision : 0.8660867453574451
recall : 0.865988909426987
F1 Score : 0.8660971494408414


#Parameters for Hyper Parameter Tuning
lr_param={"C":np.logspace(-3,3,7), "penalty":["l1","l2"],}# l1 lasso l2 ridge
param_comb = 100


#Hyper parameter Tuning of Logistic regression
from sklearn.model_selection import RandomizedSearchCV
random_search = RandomizedSearchCV(lr, param_distributions=lr_param, n_iter=param_comb, scoring='accuracy', n_jobs= -1, verbose=3, random_state=10 )


random_search.fit(X_train, y_train)

C:\Users\91989\anaconda\lib\site-packages\sklearn\model_selection\_search.py:292: UserWarning:

The total space of parameters 14 is smaller than n_iter=100. Running 14 iterations. For exhaustive searches, use GridSearchCV.

Fitting 5 folds for each of 14 candidates, totalling 70 fits

C:\Users\91989\anaconda\lib\site-packages\sklearn\model_selection\_validation.py:372: FitFailedWarning:


35 fits failed out of a total of 70.
The score on these train-test partitions for these parameters will be set to nan.
If these failures are not expected, you can try to debug them by setting error_score='raise'.

Below are more details about the failures:
--------------------------------------------------------------------------------
35 fits failed with the following error:
Traceback (most recent call last):
  File "C:\Users\91989\anaconda\lib\site-packages\sklearn\model_selection\_validation.py", line 681, in _fit_and_score
    estimator.fit(X_train, y_train, **fit_params)
  File "C:\Users\91989\anaconda\lib\site-packages\sklearn\linear_model\_logistic.py", line 1461, in fit
    solver = _check_solver(self.solver, self.penalty, self.dual)
  File "C:\Users\91989\anaconda\lib\site-packages\sklearn\linear_model\_logistic.py", line 447, in _check_solver
    raise ValueError(
ValueError: Solver lbfgs supports only 'l2' or 'none' penalties, got l1 penalty.


C:\Users\91989\anaconda\lib\site-packages\sklearn\model_selection\_search.py:969: UserWarning:

One or more of the test scores are non-finite: [       nan 0.87270336        nan 0.89212563        nan 0.89769242
        nan 0.89818746        nan 0.89800201        nan 0.89794013
        nan 0.89781641]

C:\Users\91989\anaconda\lib\site-packages\sklearn\linear_model\_logistic.py:814: ConvergenceWarning:

lbfgs failed to converge (status=1):
STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.

Increase the number of iterations (max_iter) or scale the data as shown in:
    https://scikit-learn.org/stable/modules/preprocessing.html
Please also refer to the documentation for alternative solver options:
    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression

RandomizedSearchCV(estimator=LogisticRegression(C=0.01), n_iter=100, n_jobs=-1,
                   param_distributions={'C': array([1.e-03, 1.e-02, 1.e-01, 1.e+00, 1.e+01, 1.e+02, 1.e+03]),
                                        'penalty': ['l1', 'l2']},
                   random_state=10, scoring='accuracy', verbose=3)


print('\n All results:')
print(random_search.cv_results_)
print('\n Best estimator:')
print(random_search.best_estimator_)
print('\n Best hyperparameters:')
print(random_search.best_params_)
results = pd.DataFrame(random_search.cv_results_)

 All results:
{'mean_fit_time': array([0.00800533, 0.45098314, 0.01099806, 0.8821537 , 0.00872512,
       1.12705445, 0.00943842, 1.1424911 , 0.00910869, 1.15644922,
       0.0103354 , 1.10911641, 0.01041908, 0.91872277]), 'std_fit_time': array([0.00108684, 0.02280475, 0.00316146, 0.04443592, 0.00162511,
       0.04627641, 0.00098607, 0.04112939, 0.00092771, 0.03340425,
       0.00096607, 0.05986003, 0.00104655, 0.07881173]), 'mean_score_time': array([0.        , 0.00541573, 0.        , 0.00510416, 0.        ,
       0.00580802, 0.        , 0.00542779, 0.        , 0.00611739,
       0.        , 0.00633936, 0.        , 0.00379386]), 'std_score_time': array([0.        , 0.00037446, 0.        , 0.00052716, 0.        ,
       0.00097798, 0.        , 0.00050467, 0.        , 0.0003661 ,
       0.        , 0.00093869, 0.        , 0.00074583]), 'param_penalty': masked_array(data=['l1', 'l2', 'l1', 'l2', 'l1', 'l2', 'l1', 'l2', 'l1',
                   'l2', 'l1', 'l2', 'l1', 'l2'],
             mask=[False, False, False, False, False, False, False, False,
                   False, False, False, False, False, False],
       fill_value='?',
            dtype=object), 'param_C': masked_array(data=[0.001, 0.001, 0.01, 0.01, 0.1, 0.1, 1.0, 1.0, 10.0,
                   10.0, 100.0, 100.0, 1000.0, 1000.0],
             mask=[False, False, False, False, False, False, False, False,
                   False, False, False, False, False, False],
       fill_value='?',
            dtype=object), 'params': [{'penalty': 'l1', 'C': 0.001}, {'penalty': 'l2', 'C': 0.001}, {'penalty': 'l1', 'C': 0.01}, {'penalty': 'l2', 'C': 0.01}, {'penalty': 'l1', 'C': 0.1}, {'penalty': 'l2', 'C': 0.1}, {'penalty': 'l1', 'C': 1.0}, {'penalty': 'l2', 'C': 1.0}, {'penalty': 'l1', 'C': 10.0}, {'penalty': 'l2', 'C': 10.0}, {'penalty': 'l1', 'C': 100.0}, {'penalty': 'l2', 'C': 100.0}, {'penalty': 'l1', 'C': 1000.0}, {'penalty': 'l2', 'C': 1000.0}], 'split0_test_score': array([       nan, 0.87940631,        nan, 0.90074212,        nan,
       0.90290662,        nan, 0.90352505,        nan, 0.90136054,
              nan, 0.90228819,        nan, 0.90197897]), 'split1_test_score': array([       nan, 0.8707483 ,        nan, 0.88837353,        nan,
       0.89919604,        nan, 0.8961039 ,        nan, 0.8961039 ,
              nan, 0.89548547,        nan, 0.89579468]), 'split2_test_score': array([       nan, 0.86019177,        nan, 0.88277142,        nan,
       0.89050418,        nan, 0.89236004,        nan, 0.89297866,
              nan, 0.89205073,        nan, 0.89205073]), 'split3_test_score': array([       nan, 0.87596659,        nan, 0.89205073,        nan,
       0.89328797,        nan, 0.8942159 ,        nan, 0.89483452,
              nan, 0.89514383,        nan, 0.89483452]), 'split4_test_score': array([       nan, 0.87720384,        nan, 0.89669038,        nan,
       0.90256727,        nan, 0.90473245,        nan, 0.90473245,
              nan, 0.90473245,        nan, 0.90442314]), 'mean_test_score': array([       nan, 0.87270336,        nan, 0.89212563,        nan,
       0.89769242,        nan, 0.89818746,        nan, 0.89800201,
              nan, 0.89794013,        nan, 0.89781641]), 'std_test_score': array([       nan, 0.00687306,        nan, 0.00627162,        nan,
       0.00498564,        nan, 0.005008  ,        nan, 0.00436917,
              nan, 0.00476594,        nan, 0.00463029]), 'rank_test_score': array([ 8,  7,  9,  6, 10,  5, 11,  1, 12,  2, 13,  3, 14,  4])}

 Best estimator:
LogisticRegression()

 Best hyperparameters:
{'penalty': 'l2', 'C': 1.0}


from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import KFold
from sklearn.model_selection import cross_val_score
tuned_lr = LogisticRegression(penalty = 'l2', C = 1.0 )
tuned_lr.fit(X_train, y_train)
y_pred = tuned_lr.predict(X_test)
#10-fold cross validation score
cv = cross_val_score(estimator = lr, X = X_train, y = y_train, cv =3)
tuned_log_cv = cv.mean()
print("10 fold cross validation :", log_cv)
from sklearn.metrics import accuracy_score
tuned_log_ac = accuracy_score(y_pred,y_test)
print("accuracy :", log_ac)
from sklearn.metrics import precision_score
tuned_log_p = precision_score(y_pred,y_test,average='weighted')
print("precision :", log_p)
from sklearn.metrics import recall_score
tuned_log_r = recall_score(y_pred,y_test,average='weighted')
print("recall :", log_r)
from sklearn.metrics import f1_score
tuned_log_f1 = f1_score(y_test, y_pred,average='weighted')
print("F1 Score :", log_f1)

C:\Users\91989\anaconda\lib\site-packages\sklearn\linear_model\_logistic.py:814: ConvergenceWarning:

lbfgs failed to converge (status=1):
STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.

Increase the number of iterations (max_iter) or scale the data as shown in:
    https://scikit-learn.org/stable/modules/preprocessing.html
Please also refer to the documentation for alternative solver options:
    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression

10 fold cross validation : 0.8915073916001731
accuracy : 0.865988909426987
precision : 0.8660867453574451
recall : 0.865988909426987
F1 Score : 0.8660971494408414


from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import KFold
from sklearn.model_selection import cross_val_score
rf = RandomForestClassifier(n_estimators = 1860, min_samples_split = 2, min_samples_leaf = 4, max_features='auto', max_depth = 70, bootstrap = 'True')
rf.fit(X_train, y_train)
# Predicting the Test set results
y_pred = rf.predict(X_test)
#10-fold cross validation score
cv = cross_val_score(estimator = rf, X = X_train, y = y_train, cv =3)
rf_cv = cv.mean()
print("10 fold cross validation :", rf_cv)
from sklearn.metrics import accuracy_score
rf_ac = accuracy_score(y_pred,y_test)
print("accuracy :", rf_ac)
from sklearn.metrics import precision_score
rf_p = precision_score(y_pred,y_test,average='weighted')
print("precision :", rf_p)
from sklearn.metrics import recall_score
rf_r = recall_score(y_pred,y_test,average='weighted')
print("recall :", rf_r)
from sklearn.metrics import f1_score
rf_f1 = f1_score(y_test, y_pred,average='weighted')
print("F1 Score :", rf_f1)

10 fold cross validation : 0.901094822787159
accuracy : 0.8739371534195933
precision : 0.8749279444562226
recall : 0.8739371534195933
F1 Score : 0.8738406313696206


from sklearn.model_selection import RandomizedSearchCV
max_features = ['auto', 'sqrt']
n_estimators = range(200, 2000, 10)
max_depth = range(10, 110, 10)
min_samples_split = range(2, 10, 1)
min_samples_leaf = [1, 2, 4]
bootstrap = [True, False]
random_grid = {'n_estimators': n_estimators,
               'max_features': max_features,
               'max_depth': max_depth,
               'min_samples_split': min_samples_split,
               'min_samples_leaf': min_samples_leaf,
               'bootstrap': bootstrap}
param_comb = 100


random_search = RandomizedSearchCV(estimator = rf, param_distributions = random_grid, n_iter = 10, cv = 2, verbose=2, random_state=42, n_jobs = -1)
random_search.fit(X_train, y_train)

Fitting 2 folds for each of 10 candidates, totalling 20 fits

RandomizedSearchCV(cv=2,
                   estimator=RandomForestClassifier(bootstrap='True',
                                                    max_depth=70,
                                                    min_samples_leaf=4,
                                                    n_estimators=1860),
                   n_jobs=-1,
                   param_distributions={'bootstrap': [True, False],
                                        'max_depth': range(10, 110, 10),
                                        'max_features': ['auto', 'sqrt'],
                                        'min_samples_leaf': [1, 2, 4],
                                        'min_samples_split': range(2, 10),
                                        'n_estimators': range(200, 2000, 10)},
                   random_state=42, verbose=2)


print('\n All results:')
print(random_search.cv_results_)
print('\n Best estimator:')
print(random_search.best_estimator_)
print('\n Best hyperparameters:')
print(random_search.best_params_)
results = pd.DataFrame(random_search.cv_results_)

 All results:
{'mean_fit_time': array([ 59.00887847,  94.96244752, 102.71881282,  15.7013253 ,
       105.37056756,  69.40369427,  61.39910555, 105.86983871,
        89.3273015 ,  16.00332642]), 'std_fit_time': array([2.15685344, 1.72794354, 0.31142914, 0.34644377, 2.51075745,
       1.27985775, 1.83904052, 3.47259951, 1.15950322, 0.18060207]), 'mean_score_time': array([4.63433254, 7.65199649, 9.01930237, 1.03971982, 9.75910258,
       4.83634305, 5.018134  , 9.88080311, 8.24152803, 1.00658846]), 'std_score_time': array([0.08076584, 0.38120592, 0.48870373, 0.02419591, 0.25999403,
       0.22495484, 0.00219953, 0.02264977, 0.58476591, 0.02843165]), 'param_n_estimators': masked_array(data=[1180, 1870, 1920, 340, 1990, 1280, 1860, 1970, 1660,
                   380],
             mask=[False, False, False, False, False, False, False, False,
                   False, False],
       fill_value='?',
            dtype=object), 'param_min_samples_split': masked_array(data=[7, 9, 6, 2, 3, 6, 2, 4, 8, 8],
             mask=[False, False, False, False, False, False, False, False,
                   False, False],
       fill_value='?',
            dtype=object), 'param_min_samples_leaf': masked_array(data=[1, 4, 2, 1, 4, 2, 4, 4, 4, 1],
             mask=[False, False, False, False, False, False, False, False,
                   False, False],
       fill_value='?',
            dtype=object), 'param_max_features': masked_array(data=['auto', 'sqrt', 'auto', 'auto', 'sqrt', 'sqrt', 'auto',
                   'sqrt', 'auto', 'auto'],
             mask=[False, False, False, False, False, False, False, False,
                   False, False],
       fill_value='?',
            dtype=object), 'param_max_depth': masked_array(data=[50, 70, 60, 30, 40, 30, 70, 60, 100, 10],
             mask=[False, False, False, False, False, False, False, False,
                   False, False],
       fill_value='?',
            dtype=object), 'param_bootstrap': masked_array(data=[False, False, False, False, False, False, True, False,
                   False, False],
             mask=[False, False, False, False, False, False, False, False,
                   False, False],
       fill_value='?',
            dtype=object), 'params': [{'n_estimators': 1180, 'min_samples_split': 7, 'min_samples_leaf': 1, 'max_features': 'auto', 'max_depth': 50, 'bootstrap': False}, {'n_estimators': 1870, 'min_samples_split': 9, 'min_samples_leaf': 4, 'max_features': 'sqrt', 'max_depth': 70, 'bootstrap': False}, {'n_estimators': 1920, 'min_samples_split': 6, 'min_samples_leaf': 2, 'max_features': 'auto', 'max_depth': 60, 'bootstrap': False}, {'n_estimators': 340, 'min_samples_split': 2, 'min_samples_leaf': 1, 'max_features': 'auto', 'max_depth': 30, 'bootstrap': False}, {'n_estimators': 1990, 'min_samples_split': 3, 'min_samples_leaf': 4, 'max_features': 'sqrt', 'max_depth': 40, 'bootstrap': False}, {'n_estimators': 1280, 'min_samples_split': 6, 'min_samples_leaf': 2, 'max_features': 'sqrt', 'max_depth': 30, 'bootstrap': False}, {'n_estimators': 1860, 'min_samples_split': 2, 'min_samples_leaf': 4, 'max_features': 'auto', 'max_depth': 70, 'bootstrap': True}, {'n_estimators': 1970, 'min_samples_split': 4, 'min_samples_leaf': 4, 'max_features': 'sqrt', 'max_depth': 60, 'bootstrap': False}, {'n_estimators': 1660, 'min_samples_split': 8, 'min_samples_leaf': 4, 'max_features': 'auto', 'max_depth': 100, 'bootstrap': False}, {'n_estimators': 380, 'min_samples_split': 8, 'min_samples_leaf': 1, 'max_features': 'auto', 'max_depth': 10, 'bootstrap': False}], 'split0_test_score': array([0.89930727, 0.89807026, 0.89831766, 0.89955468, 0.89794656,
       0.89831766, 0.89683325, 0.89819396, 0.89745176, 0.89398812]), 'split1_test_score': array([0.90337746, 0.90214029, 0.90387232, 0.90337746, 0.90300631,
       0.90337746, 0.90201658, 0.90238773, 0.90313003, 0.89879995]), 'mean_test_score': array([0.90134237, 0.90010528, 0.90109499, 0.90146607, 0.90047644,
       0.90084756, 0.89942491, 0.90029085, 0.90029089, 0.89639404]), 'std_test_score': array([0.00203509, 0.00203502, 0.00277733, 0.00191139, 0.00252987,
       0.0025299 , 0.00259166, 0.00209688, 0.00283913, 0.00240591]), 'rank_test_score': array([ 2,  8,  3,  1,  5,  4,  9,  7,  6, 10])}

 Best estimator:
RandomForestClassifier(bootstrap=False, max_depth=30, n_estimators=340)

 Best hyperparameters:
{'n_estimators': 340, 'min_samples_split': 2, 'min_samples_leaf': 1, 'max_features': 'auto', 'max_depth': 30, 'bootstrap': False}


from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import KFold
from sklearn.model_selection import cross_val_score
tuned_rf = RandomForestClassifier(n_estimators = 340, min_samples_split = 2, min_samples_leaf = 1, max_features='auto', max_depth = 30, bootstrap = 'False')
tuned_rf.fit(X_train, y_train)
# Predicting the Test set results
y_pred = tuned_rf.predict(X_test)
#10-fold cross validation score
cv = cross_val_score(estimator = rf, X = X_train, y = y_train, cv =3)
tuned_rf_cv = cv.mean()
print("10 fold cross validation :", rf_cv)
from sklearn.metrics import accuracy_score
tuned_rf_ac = accuracy_score(y_pred,y_test)
print("accuracy :", rf_ac)
from sklearn.metrics import precision_score
tuned_rf_p = precision_score(y_pred,y_test,average='weighted')
print("precision :", rf_p)
from sklearn.metrics import recall_score
tuned_rf_r = recall_score(y_pred,y_test,average='weighted')
print("recall :", rf_r)
from sklearn.metrics import f1_score
tuned_rf_f1 = f1_score(y_test, y_pred,average='weighted')
print("F1 Score :", rf_f1)

10 fold cross validation : 0.901094822787159
accuracy : 0.8739371534195933
precision : 0.8749279444562226
recall : 0.8739371534195933
F1 Score : 0.8738406313696206


# Fitting Decision Tree Classification to the Training set
from sklearn.tree import DecisionTreeClassifier
from sklearn.model_selection import KFold
from sklearn.model_selection import cross_val_score
classifier = DecisionTreeClassifier(min_samples_leaf = 6, max_features=7, max_depth= None, criterion = 'gini')
classifier.fit(X_train, y_train)

# Predicting the Test set results
y_pred = classifier.predict(X_test)

#10-fold cross validation score
cv = cross_val_score(estimator = classifier, X = X_train, y = y_train, cv =3)
dc_cv = cv.mean()
print("10 fold cross validation :", dc_cv)
from sklearn.metrics import accuracy_score
dc_ac = accuracy_score(y_pred,y_test)
print("accuracy :", dc_ac)
from sklearn.metrics import precision_score
dc_p = precision_score(y_pred,y_test,average='weighted')
print("precision :", dc_p)
from sklearn.metrics import recall_score
dc_r = recall_score(y_pred,y_test,average='weighted')
print("recall :", dc_r)
from sklearn.metrics import f1_score
dc_f1 = f1_score(y_test, y_pred,average='weighted')
print("F1 Score :", dc_f1)

10 fold cross validation : 0.8485804416403786
accuracy : 0.8083179297597043
precision : 0.8098093342521533
recall : 0.8083179297597043
F1 Score : 0.8077182156541395


param_dist = {"max_depth": [3, None],
              "max_features": range(1, 9),
              "min_samples_leaf": range(1, 9),
              "criterion": ["gini", "entropy"]}


random_search = RandomizedSearchCV(estimator = classifier, param_distributions = param_dist, n_iter = 10, cv = 5, verbose=2, random_state=42, n_jobs = -1)
random_search.fit(X_train, y_train)

Fitting 5 folds for each of 10 candidates, totalling 50 fits

RandomizedSearchCV(cv=5,
                   estimator=DecisionTreeClassifier(max_features=7,
                                                    min_samples_leaf=6),
                   n_jobs=-1,
                   param_distributions={'criterion': ['gini', 'entropy'],
                                        'max_depth': [3, None],
                                        'max_features': range(1, 9),
                                        'min_samples_leaf': range(1, 9)},
                   random_state=42, verbose=2)


print('\n All results:')
print(random_search.cv_results_)
print('\n Best estimator:')
print(random_search.best_estimator_)
print('\n Best hyperparameters:')
print(random_search.best_params_)
results = pd.DataFrame(random_search.cv_results_)

 All results:
{'mean_fit_time': array([0.09387021, 0.03995581, 0.05808277, 0.06206493, 0.09689026,
       0.06814876, 0.12480278, 0.06617355, 0.04326472, 0.04342818]), 'std_fit_time': array([0.00633166, 0.00222616, 0.00562742, 0.00916182, 0.00684135,
       0.0154164 , 0.00382658, 0.00929117, 0.00565482, 0.00473721]), 'mean_score_time': array([0.00598259, 0.00598383, 0.00615144, 0.00601416, 0.00618849,
       0.005024  , 0.00595036, 0.00544682, 0.00515532, 0.00457697]), 'std_score_time': array([1.25584364e-03, 8.98739154e-04, 4.80861821e-04, 6.38994529e-05,
       9.83500753e-04, 4.16537562e-04, 5.33382905e-04, 4.90025403e-04,
       2.11352120e-04, 9.33533948e-04]), 'param_min_samples_leaf': masked_array(data=[5, 7, 8, 7, 6, 2, 3, 8, 2, 7],
             mask=[False, False, False, False, False, False, False, False,
                   False, False],
       fill_value='?',
            dtype=object), 'param_max_features': masked_array(data=[5, 1, 2, 2, 7, 8, 7, 5, 2, 4],
             mask=[False, False, False, False, False, False, False, False,
                   False, False],
       fill_value='?',
            dtype=object), 'param_max_depth': masked_array(data=[None, 3, None, None, None, 3, None, 3, 3, 3],
             mask=[False, False, False, False, False, False, False, False,
                   False, False],
       fill_value='?',
            dtype=object), 'param_criterion': masked_array(data=['entropy', 'gini', 'gini', 'entropy', 'gini',
                   'entropy', 'entropy', 'entropy', 'gini', 'gini'],
             mask=[False, False, False, False, False, False, False, False,
                   False, False],
       fill_value='?',
            dtype=object), 'params': [{'min_samples_leaf': 5, 'max_features': 5, 'max_depth': None, 'criterion': 'entropy'}, {'min_samples_leaf': 7, 'max_features': 1, 'max_depth': 3, 'criterion': 'gini'}, {'min_samples_leaf': 8, 'max_features': 2, 'max_depth': None, 'criterion': 'gini'}, {'min_samples_leaf': 7, 'max_features': 2, 'max_depth': None, 'criterion': 'entropy'}, {'min_samples_leaf': 6, 'max_features': 7, 'max_depth': None, 'criterion': 'gini'}, {'min_samples_leaf': 2, 'max_features': 8, 'max_depth': 3, 'criterion': 'entropy'}, {'min_samples_leaf': 3, 'max_features': 7, 'max_depth': None, 'criterion': 'entropy'}, {'min_samples_leaf': 8, 'max_features': 5, 'max_depth': 3, 'criterion': 'entropy'}, {'min_samples_leaf': 2, 'max_features': 2, 'max_depth': 3, 'criterion': 'gini'}, {'min_samples_leaf': 7, 'max_features': 4, 'max_depth': 3, 'criterion': 'gini'}], 'split0_test_score': array([0.8475572 , 0.70655535, 0.82529375, 0.83147805, 0.84632035,
       0.75881262, 0.84353741, 0.70810142, 0.73809524, 0.74737168]), 'split1_test_score': array([0.84972171, 0.70562771, 0.82560297, 0.83920841, 0.84910328,
       0.71428571, 0.83426098, 0.7563389 , 0.75170068, 0.76777984]), 'split2_test_score': array([0.83854006, 0.68017321, 0.83544695, 0.82245592, 0.83544695,
       0.8097742 , 0.85029384, 0.75069595, 0.73182802, 0.71326941]), 'split3_test_score': array([0.84658212, 0.67553356, 0.83823075, 0.81626972, 0.85709867,
       0.7884318 , 0.84936591, 0.7915249 , 0.69007114, 0.73430251]), 'split4_test_score': array([0.84441695, 0.64924219, 0.82709558, 0.82554903, 0.85740798,
       0.73430251, 0.85524281, 0.74822147, 0.7339932 , 0.77729663]), 'mean_test_score': array([0.84536361, 0.6834264 , 0.830334  , 0.82699222, 0.84907545,
       0.76112137, 0.84654019, 0.75097653, 0.72913766, 0.74800401]), 'std_test_score': array([0.00381436, 0.02130393, 0.00541803, 0.0078357 , 0.00808864,
       0.03473073, 0.00717783, 0.02652874, 0.02071876, 0.02299463]), 'rank_test_score': array([ 3, 10,  4,  5,  1,  6,  2,  7,  9,  8])}

 Best estimator:
DecisionTreeClassifier(max_features=7, min_samples_leaf=6)

 Best hyperparameters:
{'min_samples_leaf': 6, 'max_features': 7, 'max_depth': None, 'criterion': 'gini'}


# Fitting Decision Tree Classification to the Training set
from sklearn.tree import DecisionTreeClassifier
from sklearn.model_selection import KFold
from sklearn.model_selection import cross_val_score
tuned_classifier = DecisionTreeClassifier(min_samples_leaf = 6, max_features=7, max_depth= None, criterion = 'gini')
tuned_classifier.fit(X_train, y_train)

# Predicting the Test set results
y_pred = tuned_classifier.predict(X_test)

#10-fold cross validation score
cv = cross_val_score(estimator = classifier, X = X_train, y = y_train, cv =3)
tuned_dc_cv = cv.mean()
print("10 fold cross validation :", dc_cv)
from sklearn.metrics import accuracy_score
tuned_dc_ac = accuracy_score(y_pred,y_test)
print("accuracy :", dc_ac)
from sklearn.metrics import precision_score
tuned_dc_p = precision_score(y_pred,y_test,average='weighted')
print("precision :", dc_p)
from sklearn.metrics import recall_score
tuned_dc_r = recall_score(y_pred,y_test,average='weighted')
print("recall :", dc_r)
from sklearn.metrics import f1_score
tuned_dc_f1 = f1_score(y_test, y_pred,average='weighted')
print("F1 Score :", dc_f1)

10 fold cross validation : 0.8485804416403786
accuracy : 0.8083179297597043
precision : 0.8098093342521533
recall : 0.8083179297597043
F1 Score : 0.8077182156541395


def eval_result(rf, X_test, y_test, validation = False):
    with warnings.catch_warnings():
        warnings.simplefilter("ignore")
        pred = rf.predict(X_test)
        print(classification_report(y_test, pred, target_names = ['Forward', 'Midfielder', 'Defender', 'Goalkeeper']))
        display(pd.DataFrame(confusion_matrix(y_test, pred), 
                         columns = ['Predicted Forward', 'Predicted Midfielder', 'Predicted Defender', 'Predicted Goalkeeper'],
                         index = ['Forward', 'Midfielder', 'Defender', 'Goalkeeper']))
        
        if hasattr(rf, 'feature_importances_') and not validation:
            features = pd.DataFrame({
                'Variable'  :X_test.columns,
                'Importance':rf.feature_importances_
            })
            features.sort_values('Importance', ascending=False, inplace=True)
            display(features.head(20))


eval_result(rf, X_test, y_test, validation = True)

              precision    recall  f1-score   support

     Forward       0.92      0.92      0.92      1713
  Midfielder       0.85      0.79      0.82      1153
    Defender       1.00      1.00      1.00       505
  Goalkeeper       0.82      0.86      0.84      2039

    accuracy                           0.87      5410
   macro avg       0.90      0.89      0.89      5410
weighted avg       0.87      0.87      0.87      5410


# displaying confusion matrix and heat map to find correlation
cm = confusion_matrix(y_test,rf.predict(X_test))
sns.heatmap(cm,annot=True,fmt="d")

<AxesSubplot:>


def eval_result(tuned_rf, X_test, y_test, validation = False):
    with warnings.catch_warnings():
        warnings.simplefilter("ignore")
        pred = tuned_rf.predict(X_test)
        print(classification_report(y_test, pred, target_names = ['Forward', 'Midfielder', 'Defender', 'Goalkeeper']))
        display(pd.DataFrame(confusion_matrix(y_test, pred), 
                         columns = ['Predicted Forward', 'Predicted Midfielder', 'Predicted Defender', 'Predicted Goalkeeper'],
                         index = ['Forward', 'Midfielder', 'Defender', 'Goalkeeper']))
        
        if hasattr(tuned_rf, 'feature_importances_') and not validation:
            features = pd.DataFrame({
                'Variable'  :X_test.columns,
                'Importance':tuned_rf.feature_importances_
            })
            features.sort_values('Importance', ascending=False, inplace=True)
            display(features.head(20))


eval_result(tuned_rf, X_test, y_test, validation = True)

              precision    recall  f1-score   support

     Forward       0.92      0.92      0.92      1713
  Midfielder       0.85      0.79      0.82      1153
    Defender       1.00      1.00      1.00       505
  Goalkeeper       0.82      0.85      0.84      2039

    accuracy                           0.87      5410
   macro avg       0.90      0.89      0.89      5410
weighted avg       0.87      0.87      0.87      5410


# displaying confusion matrix and heat map to find correlation
cm = confusion_matrix(y_test,tuned_rf.predict(X_test))
sns.heatmap(cm,annot=True,fmt="d")

<AxesSubplot:>


accuracy = [SVC_ac,log_ac,rf_ac,dc_ac]
ten_fold_cross_validation = [SVC_cv,log_cv,rf_cv,dc_cv]
f1_score = [SVC_f1,log_f1,rf_f1,dc_f1]
precision = [SVC_p,log_p,rf_p,dc_p]
recall = [SVC_r,log_r,rf_r,dc_r]
models = DataFrame({'Accuracy': accuracy, '10 fold cv': ten_fold_cross_validation, 
                    "f1 score": f1_score,'precision': precision,
                   'recall': recall})
models.index = ['Support Vector Classifier','Logistic Regression Classifier','Random Forest Classifier','Decision Tree Classifier']
models


models1 = DataFrame({'Accuracy' : models.unstack()}).reset_index()
# plot accuracies
plt.figure(figsize=(8, 7))
fig_models = sns.barplot(x='level_0', y='Accuracy', hue='level_1', data=models1);
fig_models.set(xlabel='Evaluation Metrics', ylabel='Evaluation Metrics Value');
fig_models.set_title('The Accuracy of All Ensemble Models Over Five Evaluation Metrics');


accuracy = [tuned_SVC_ac,tuned_log_ac,tuned_rf_ac,tuned_dc_ac]
ten_fold_cross_validation = [tuned_SVC_cv,tuned_log_cv,tuned_rf_cv,tuned_dc_cv]
f1_score = [tuned_SVC_f1,tuned_log_f1,tuned_rf_f1,tuned_dc_f1]
precision = [tuned_SVC_p,tuned_log_p,tuned_rf_p,tuned_dc_p]
recall = [tuned_SVC_r,tuned_log_r,tuned_rf_r,tuned_dc_r]
models = DataFrame({'Accuracy': accuracy, '10 fold cv': ten_fold_cross_validation, 
                    "f1 score": f1_score,'precision': precision,
                   'recall': recall})
models.index = ['Support Vector Classifier','Logistic Regression Classifier','Random Forest Classifier','Decision Tree Classifier']
models


models1 = DataFrame({'Accuracy' : models.unstack()}).reset_index()
# plot accuracies
plt.figure(figsize=(8, 7))
fig_models = sns.barplot(x='level_0', y='Accuracy', hue='level_1', data=models1);
fig_models.set(xlabel='Evaluation Metrics', ylabel='Evaluation Metrics Value');
fig_models.set_title('The Accuracy of All Ensemble Models Over Five Evaluation Metrics');

	Accuracy	10 fold cv	f1 score	precision	recall
Support Vector Classifier	0.868022	0.890951	0.868052	0.868572	0.868022
Logistic Regression Classifier	0.867837	0.891507	0.867936	0.867925	0.867837
Random Forest Classifier	0.873937	0.900786	0.873835	0.874838	0.873937
Decision Tree Classifier	0.810351	0.838498	0.810158	0.810781	0.810351

Importing Libraries to build Ensemble Machine Learning Algorithms¶

File Reading¶

Data Pre-Processing¶

Checking the distributions, skewness and Inter Quartile Range to fill the null values in the data¶

Filling the null values with mean after checking the distributions¶

Exploratory Data Analysis¶

Feature Selection¶

Dataframe creation after feature selection¶

Model Building¶

Support Vector Classifier¶

Tuned Model for SVC¶

Logistic Regression Classifier¶

Tuned Model for Logistic Regression Classifier¶

Random Forest Classifier¶

Decision Tree Classifier¶

Random Forest Classifier Base Model Evaluation Metrics¶

Random Forest Classifier Base Model Confusion Matrix¶

Random Forest Classifier Base Model Heat Map¶

Random Forest Classifier Tuned Model Evaluation Metrics¶

Random Forest Classifier Tuned Model Confusion Matrix¶

Random Forest Classifier Tuned Model Heat Map¶

Model Summary¶

Model Summary for 4 Major Positions Classifier Base Model¶

Model Summary for 4 Major Positions Classifier Tuned Model¶

Random Forest Classifier performs better with above Both Base and Tuned models evaluation metrics when compared to all other 3 models in predicting the major 4 position of the player¶

	Unnamed: 0	ID	Name	Age	Photo	Nationality	Flag	Overall	Potential	Club	...	Penalties	Composure	Defensive Awareness	Standing Tackle	Sliding Tackle	GK Diving	GK Handling	GK Kicking	GK Positioning	GK Reflexes
0	0	253283	Facundo Pellistri	18	https://cdn.sofifa.com/players/253/283/20_60.png	Uruguay	https://cdn.sofifa.com/flags/uy.png	71	87	Peñarol	...	66.0	61.0	35.0	11.0	18.0	9.0	12.0	7.0	8.0	7.0
1	1	179813	Edinson Cavani	32	https://cdn.sofifa.com/players/179/813/20_60.png	Uruguay	https://cdn.sofifa.com/flags/uy.png	86	86	Paris Saint-Germain	...	85.0	80.0	57.0	48.0	39.0	12.0	5.0	13.0	13.0	10.0
2	2	245541	Giovanni Reyna	17	https://cdn.sofifa.com/players/245/541/20_60.png	United States	https://cdn.sofifa.com/flags/us.png	68	87	Borussia Dortmund	...	50.0	59.0	30.0	23.0	24.0	10.0	13.0	14.0	12.0	7.0
3	3	233419	Raphael Dias Belloli	23	https://cdn.sofifa.com/players/233/419/20_60.png	Brazil	https://cdn.sofifa.com/flags/br.png	81	85	Stade Rennais FC	...	73.0	79.0	45.0	54.0	38.0	8.0	7.0	13.0	8.0	14.0
4	4	198710	James Rodríguez	28	https://cdn.sofifa.com/players/198/710/20_60.png	Colombia	https://cdn.sofifa.com/flags/co.png	82	82	Everton	...	81.0	87.0	52.0	41.0	44.0	15.0	15.0	15.0	5.0	14.0

	Name	Club	Age	Position	Potential
0	Florian Wirtz	Bayer 04 Leverkusen	17	RCM	88
1	Isak Jansson	Kalmar FF	18	RF	73
2	Nuno Alexandre Tavares Mendes	Sporting CP	18	LWB	87
3	Reinier Jesus Carvalho	Borussia Dortmund	18	CF	87
4	Oliver Skipp	Norwich City	19	RDM	84
5	Jeremie Frimpong	Celtic	19	RWB	86
6	Thiago Almada	Vélez Sarsfield	19	LCM	89
7	Houssem Aouar	Olympique Lyonnais	22	LCM	89
8	Alphonso Davies	FC Bayern München	19	LB	89
9	Sergio Reguilón Rodríguez	Tottenham Hotspur	23	LB	89
10	Dominik Szoboszlai	FC Red Bull Salzburg	19	LAM	87
11	Vinícius José de Oliveira Júnior	Real Madrid	19	LW	93
12	Matthijs de Ligt	Juventus	20	CB	92
13	João Félix Sequeira	Atlético Madrid	20	RS	93
14	Jadon Sancho	Borussia Dortmund	20	RW	93
15	Antony Matheus dos Santos	Ajax	20	RAM	88
16	Sandro Tonali	Milan	20	CDM	91
17	Nils Fröling	Kalmar FF	20	LF	79
18	Patson Daka	FC Red Bull Salzburg	21	LS	85
19	Maximiliano Gómez	Valencia CF	23	LS	85
20	Moussa Dembélé	Olympique Lyonnais	23	LS	85
21	Emerson Leite De Souza	Real Betis	21	RB	88
22	Aaron Wan-Bissaka	Manchester United	22	RB	88
23	Kai Havertz	Chelsea	21	CAM	93
24	Edmond Tapsoba	Bayer 04 Leverkusen	21	RCB	88
25	Joe Gomez	Liverpool	23	RCB	88
26	Kylian Mbappé	Paris Saint-Germain	21	RM	95
27	Federico Valverde	Real Madrid	21	CM	90
28	Gianluigi Donnarumma	Milan	21	GK	92
29	Lautaro Martínez	Inter	22	ST	91
30	Marcus Rashford	Manchester United	22	LM	91
31	Frenkie de Jong	FC Barcelona	23	LDM	90
32	Niklas Süle	FC Bayern München	24	LCB	89

	Name	Age	Position	Overall
0	Remil Nadkarni	34	CAM	59
1	Anuvinda Khurana	27	CB	59
2	Bismeet Sidhu	32	CDM	60
3	Gajodara Chatterjee	34	GK	64
4	Adit Ginti	26	LB	61
5	Abhimoda Chakraborty	34	LCB	61
6	Chapal Palan	29	LCM	57
7	Attana Deshpande	39	LM	60
8	Halicharan Narzary	24	LM	60
9	Sunil Chhetri	33	LS	67
10	Lalrindika Ralte	25	LW	61
11	Bhadrashree Raj	32	RB	64
12	Tapish Atwal	36	RCB	59
13	Anvit Swaminathan	28	RCM	61
14	Hantidev Bhandari	31	RM	62
15	Jeje Lalpekhlua	27	ST	63

	Predicted Forward	Predicted Midfielder	Predicted Defender	Predicted Goalkeeper
Forward	1568	0	0	145
Midfielder	4	909	0	240
Defender	0	0	505	0
Goalkeeper	134	159	0	1746

	Predicted Forward	Predicted Midfielder	Predicted Defender	Predicted Goalkeeper
Forward	1569	0	0	144
Midfielder	5	911	0	237
Defender	0	0	505	0
Goalkeeper	134	162	0	1743

	Accuracy	10 fold cv	f1 score	precision	recall
Support Vector Classifier	0.860444	0.890951	0.860279	0.861062	0.860444
Logistic Regression Classifier	0.865989	0.891507	0.866097	0.866087	0.865989
Random Forest Classifier	0.873937	0.901095	0.873841	0.874928	0.873937
Decision Tree Classifier	0.808318	0.848580	0.807718	0.809809	0.808318